Cognizant 20-20 Insights

Employing Analytics to Automate and

Optimize Insurance Distribution
An evolving distribution landscape creates the opportunity for carriers
to use advanced analytics tools and methodologies to capitalize on
their data and extend their reach to digitally-connected consumers.
Executive Summary
For years, insurance distribution remained largely
unchanged relying primarily on independently
affiliated agents and brokers. The process of selling to and servicing insurance customers was for
the most part based on high-touch customer
encounters. These interactions consumed a lot
of time and energy requiring agents to educate
and assist customers with their research and subsequent purchase of coverage for insurable risks.
Today, new distribution channels continue
to emerge. One example is Bancassurance
among the fastest-growing online conduits for
fulfilling insurance customers needs throughout
developed and maturing markets.1 As these
new entities take center stage, the distribution
side of the carrier business faces the dual
challenge of managing growing support costs
to serve a limited (and shrinking) number of
advisors, while meeting the escalating demands
of multichannel sales and distribution.
Historically, carriers have reacted slowly to such
developments, with a modest appetite for investing in future-focused changes. This evolution vs.

cognizant 20-20 insights | april 2015

revolution mindset has led to the preferential

allocation of funds for supporting and managing
legacy systems, including personnel-dependent
distribution processes. By pursuing this keep the
wheels on the bus strategy, many carriers skirted
the opportunity to execute and capitalize on
projects with potentially transformative implications for the business.
In this white paper, we throw light on how insurers
can apply advanced analytics to their distribution
operations leveraging data and research
resulting from the analytics mandate across
the industry. Critical to this discussion is the
powerful construct of Code Halos, or the virtual
footprints individuals, organizations and devices
leave in our increasingly connected digital world.2
We also offer an approach for applying analytic
insights and innovations to key business activities
to help insurers enhance their competitiveness,
profitability and consumer centricity.

What Consumers Want

Prior to the mid-1990s, insurance distributors
held most of the knowledge regarding insurance products, pricing and processes requiring

customers to have the assistance of an intermediary. By the end of the last decade, consumers had
become more informed, thanks to social media,
the ability to compare service providers through
reviews posted on Web sites and blogs, and the
growing volumes of content streaming from handheld devices and other online channels. Clearly,
producers no longer wielded as much influence as
they once had.
In this increasingly digital environment, consumers have shown two thematic preferences for how
their insurance needs are met:

They want industry players to expand

their services beyond core comfort and
assurance; they want a sense of community,
and the freedom to choose from a variety of
highly personalized products. While todays
insurance carriers and distributors have an
opportunity to comply with these preferences
and achieve true market differentiation, they
must be ready to compete with services that
can stand up to a new breed of potentially disruptive entrants, such as Google and Amazon.3

Consumers expect carriers and their distribution partners to provide tools and technologies through various digital channels that
expedite the process of researching, applying
for and servicing their policies. Increasingly,
this expectation includes the use of highfunctioning mobile devices.
Using Analytics to Better Inform the Business
Several factors the evolution of information technology, demographic changes, new cultural norms
of social and sales interBy employing actions, and a hyper-conadvanced analytics, nected millennial generacarriers can take tion are driving dramatic
changes in the insurance
advantage of distribution ecosystem.

real-time data to
gain deeper insights
into their customers
preferences and
better inform
their business
and distribution
strategies.

We believe that carriers

that are well informed on all
fronts are in the best position to optimize available
resources, make the most
of their existing IT investments, and implement new
systems as needed to grow
the business. By employing
advanced analytics, carriers can take advantage of real-time data to gain
deeper insights into their customers preferences

cognizant 20-20 insights

and better inform their business and distribution

strategies. Yet to achieve this kind of operating
environment, carriers must first adopt a strategic, proactive approach to leveraging analytics
across the enterprise. A significant number of
insurers are already doing so investing in data
and analytics tools for marketing, distribution
and policy servicing. In fact, according to a recent
report from Strategy Meets Action (SMA),4 analytics and mobile technologies are among the top
priorities for insurers this year.
For instance, more than 40% of survey
respondents said their organizations are now
using analytics to develop advanced customer
segmentation. Another 20% noted that they
are already piloting or implementing such
environments. Furthermore, 21% are using
analytics to develop a single view of customers,
while another 26% are piloting or implementing
the tools needed to do so. Analytics are used by
48% of insurers for channel/agent performance
management, and another 20% are currently
piloting or deploying these tools.
The importance of Code HaloTM thinking is
immense when applying analytics to the
distribution side of the insurance industry.
Analytics provide the catalyst for making meaning
from intersecting Code Halos affording the
opportunity for insurers and their intermediaries
to identify, communicate with and conduct
business with the most profitable clients.

The Insurance Industrys Digital

Business Challenges
One dominant theme, particularly in personal
lines of insurance, is the emergence of the
digital consumer. Typically, he or she carries a
modern smartphone, is persistently connected
in real time to work, family and social circles,
and is contextually informed by location-aware
applications. Dealing with digital consumers
with easy access to information (e.g., product
reviews) and transactional capabilities (e.g.,
digital payments) imposes massive demands on
the industrys traditional distribution models. For
instance, channel players must continually prove
their worth by adding value to sales and service
processes. This places more pressure on them to
differentiate themselves and offer distinct value
that is difficult to replicate and automate through
digital interactions and transactions.
Clearly, the evolution of the digital consumer presents unique distribution challenges. Addressing

and overcoming these hurdles requires knowledge and a deep understanding of the changing
distribution landscape, as well as a new way of
thinking in several key areas:

Product design. Shifting consumer demographics, increasingly technology-savvy customers (existing, new and prospective alike)
and the onset of new insurable risks create some special challenges for insurance
product-design and development teams.
They are expected to create well-rounded
views of customers wants and needs, process
large amounts of data, translate existing loss
information to estimate new categories of
products, and price them competitively.

Overall, they must be able to finely segment

risks, keep anti-selection at bay, optimize
predictive underwriting and drive profitability by back-feeding claims experiences
into product designs, targeting and pricing.
Simply stated, the level of agility a product
design team can achieve in understanding
user needs and translating those into competitive product propositions largely depends
on their analytical abilities.

Multi-channel distribution. Historically, various insurance distribution channels were able

to reach existing and prospective customers
by themselves (or independently), but these
channels were rarely integrated. Todays digital

Quick Take
Decoding Customer Behavior
To truly understand customer needs and the factors that drive their actions, insurers must become
familiar with the fundamentals of customer behavior, and be prepared to answer the following questions:

What and why are customers buying? What is the implication for the product development lifecycle?
Are they happy with the insurance products and services they currently own?
How long might individual customer relationships last? How can the insurer proactively mitigate the
risk of attrition?
What leads to customer churn? How does that affect profitability?
Can a distributor predict the propensity of a customer to buy a particular product? What other
products may interest that customer?

Can the insurer sense when customers signal dissatisfaction, and adjust/tailor services to their
liking in real time?

Are there leading or lagging indicators of insurance customers preferences that could be used
to optimize marketing spend?

How can they acquire and retain the most profitable customers at the lowest distribution costs?
How can they determine customers wallet share?
How can they maximize the conversion of prospects?
Is the insurer cognizant of the competition, and prepared to win over customers to gain market
share?

How does the insurer optimize its production channels and incentives that go towards premium
production?

How do current clients and ideal prospective clients wish to be engaged digitally? Does the
company have a current understanding of how to meet their needs today, as well as a complementary
long-term strategy?

While these questions may seem fundamental to a successful product and distribution strategy, the truth
is more nuanced. The broader reality is that all but the most sophisticated carriers and distributors are
using preexisting data to understand their business and address the issues cited above. By minimizing,
overlooking or simply not taking action to address this concern, carriers miss the opportunity to fully
understand and comply with their customers preferences.

cognizant 20-20 insights

consumers have a distinct preference for flexible and integrated channels that enable them
to interact via e-mail, chat, mobile, the call
center and direct communications. They have
become accustomed to and expect seamless,
multi-channel experiences like those pioneered
by retail and digital media
Analyzing channel giants such as Amazon and
probability and Google.

productivity can
make a significant
difference when it
comes to improving
overall distribution
efficiency.

Enabling and maintaining multi-channel interactions from engagement

through distribution has
value, but it also places
an additional burden on
already-stretched IT budgets. Better resource alignment across channels requires detailed
insights into each channels productivity,
revenue-wise. This helps ensure that preferred distribution channels are prioritized
and receive the necessary resources. Analyzing channel probability and productivity can
make a significant difference when it comes
to improving overall distribution efficiency.

Marketing. Another area with limited budgets

but with increasing scrutiny around its return
on investment (ROI) is marketing (advertising, to be more specific). The growing demand
for closed-loop marketing systems is an
outcome of the push to measure and
enhance the productivity of various marketing channels. Optimizing the marketing mix is
one technique that several insurance carriers
use to increase the impact of marketing
communications and boost the ROI from
multi-channel marketing initiatives. Various distribution channels often ask for more
marketing items/collaterals in an effort
to extend their customer reach. Typically,
distribution organizations do not have information on the volume of collaterals used,
since most lack the ability to track, automate
and collect data. Using advanced analytics to
understand the effectiveness of marketing
spend can have significant, positive implications helping companies to:
Develop a deeper understanding of customers and their behavior patterns.
Reduce spend on channels and collateral
materials that do not have high returns.

cognizant 20-20 insights

Improve compliance by effectively tracking

collateral items.
Realize overall productivity improvements.
When marketing teams are no longer able to
build and/or manage an ever-increasing lineup
of channels or materials, they can refine and
improve their best-of-the-best approaches.

Profitability management. Competition and

market shifts heighten pressure on breadand-butter service lines, which impacts
carriers overall profitability. Efforts to squeeze
distribution channel spending (typically via
reduced commissions and fees) signal that
producers and brokers see less returns on the
same-sized book of business. Seeking a balance between these two opposing forces to
maximize profitability requires highly accurate
analytical insights.
Measuring distribution profitability comprises
three layers: carrier standalone, distributor
standalone and joint measurement. In the two
standalone categories, each group is responsible for the independent drill-down and
analysis of their business. Too often, distributors rely on carriers to provide the underlying data, when in reality there are many sellthrough transactions that a distributor can
discreetly track to gauge profitability. A joint
commitment to disciplined analysis and information can facilitate valuable and informed
partnerships. This process not only apprises
carriers of key profitability facts; it also carries
with it broader practical implications that can
affect product design and pricing, as well as
marketing spend.

The Science of Resource Allocation:

Using Analytics to Address Key
Challenges
An important principle of data science is that
analytics is a process with distinct stages. It is not
an event, as is commonly thought. Some stages
of analytics involve the application of information technology and computational capacity (e.g.,
data processing, automated pattern discovery).
Other stages require an analysts creative understanding of business imperatives. Each data-driven
business decision is a unique challenge, with its own
combination of goals and constraints. It is therefore
important for an analytics team to sit with business
stakeholders and develop a sound understanding

of the business challenge so it can be broken

down into analytic subtasks.
Conventional wisdom suggests that while each
business problem might have some subtasks
that are unique, many (i.e., data cleansing,
staging, tabulation, etc.) are common among
business issues. This allows for the reuse of
analytic approaches and algorithms permitting
the analytics team as a whole to be much
more efficient by addressing several problems
simultaneously.
To fully understand the entire gamut of insurance
distribution-related challenges that can be
resolved through analytics, we will review the
important analytic approaches and algorithms
that are closely aligned with the tasks at hand.
Sales Conversion Analysis
In any commercial environment, selling is a stepby-step process. This certainly holds true for
insurance distribution channels. Starting from
a potential lead, the customer passes through
various stages of the selling process, including
engagement, qualification, proposal, negotiation
and deal closure. Distributors and brokers
frequently maintain a sales funnel to track the
number of customer threads in each stage and
monitor the percentage of threads that move
forward. This can help discern weaknesses in
agent operations while maximizing sales leads

and opportunities. For example, a persistent

inability to convert qualified leads into proposals
might indicate a weak link in the lead-qualification
process.
It is essential to differentiate sales-related
activities along two dimensions: productiongenerating activities (PGAs) and productionsupporting activities (PSAs). The distinction
is important, since it forces organizations to
understand the critical path
of activities and events that Understanding the
drive revenue, as opposed
discreet nature of
to those (PSAs) that merely
support the activities that both PGAs and PSAs
actually generate revenue. affords a deeper
Once businesses recognize
which PGAs are directly understanding of the
correlated with production, full sales continuum
they can focus on the PSAs
and the impediments
that drive the highest
volume of PGAs. A common to effective sales
mistake is treating PGAs conversion.
and PSAs in the same way,
which means they are tracked in the same way.
Understanding the discreet nature of both PGAs
and PSAs affords a deeper understanding of the
full sales continuum and the impediments to
effective sales conversion.
New predictive analytics techniques, such as
classification or class probability analysis, can also
be employed to improve the efficiency of sales

Quick Take
How Analytics Helped an Insurer Maximize Customer Retention
One of the worlds top-three composite insurance groups operating in 40-plus countries requested
Cognizants insight in helping to improve customer retention. In several countries, in its life and pensions
business, the groups 13th-month persistency rates were lower than the industry norm. The company
asked us to develop a series of predictive models using logistic regression, GLM and Cox regression
techniques to correspond with observed attrition trends across time periods, external environment
changes, and the changing lifecycle stages of existing in-force customers.
Figure 1 (next page) shows how we helped improve the insurers customer-attrition challenge
using our retention and recapture analytics framework, iVALUE, which runs on a big data platform.
The predictive models within the iVALUE platform are integrated with the retention and recapture
campaign management systems through Web services.
Using analytical models, customer retention and persistency have increased substantially in multiple
countries. For one of its European implementations, the iVALUE-based solution has helped grow
customer retention from 83% (steady for the last five years) to 91% in one quarter. Moreover, the models
were also deployed in customer win-back campaigns and revived 4% of the clients lapsed/surrendered
customers. In another Asian implementation, 50% of high-risk attrition customers were retained, and
3.5% of lost customers were won back.

cognizant 20-20 insights

operations. These data-mining approaches can

be used to forecast sales quotas and conversion
probabilities. They help to answer questions
concerning the likelihood of a particular lead
proceeding to the next sales stage. This kind of

probabilistic assessment makes it easy to align

the best sales resources human and material
with leads that have the highest chances of
conversion.

L9$/8($QDO\WLFV6HUYLFHV
/HDG*HQHUDWLRQ
&RQYHUVLRQ

&XVWRPHU6HJPHQWDWLRQ
&URVV6HOO8SVHOO


6XUUHQGHU $WWULWLRQ
3UHGLFWLRQ

)UDXG 0RQH\/DXQGHULQJ
'HWHFWLRQ 

$QDO\WLFV
'HOLYHU\

0RUWDOLW\0RUELGLW\
&ODLPV3UHGLFWLRQ


5LVN &RPSOLDQFH
5HSRUWLQJ $QDO\WLFV


&ORVHG%RRN
$QDO\WLFV

'DWD3UH3URFHVVLQJ
4XDOLW\$VVHVVPHQW


9LVXDOL]DWLRQ3RUWDO
0JPW5HSRUWLQJ



&XVWRPHU 3UHPLXP
5HWHQWLRQ&RQVXOWLQJ

L9$/8(,PSOHPHQWDWLRQDQG%XVLQHVV,QWHJUDWLRQ$SSURDFK
&XVWRPL]DWLRQ &RQILJXUDWLRQ
'DWD
3UHSDUDWLRQ

3ODQQLQJ
,QLWLDWLRQ
y
'DWDDVVHVVPHQW
ZRUNVKRSZLWK
GDWDDYDLODELOLW\
FKHFNOLVW
y
'DWDVRXUFHV
LGHQWLILFDWLRQ
y
)LQDOL]H
UHTXLUHPHQWVIRU
GDWDH[WUDFWLRQ

L9$/8(/DXQFK

([SORUDWRU\
$QDO\VLV
0RGHOLQJ

0RGHO
9DOLGDWLRQ
3LORWLQJ

y
,QIUDVWUXFWXUH
VHWXS

y
L9$/8(PRGHO
WUDLQLQJ

y
'DWDPHUJLQJ
FOHDQVLQJDQG
VWDQGDUGL]DWLRQ

y
L9$/8(&XVWRPHU
6HJPHQWDWLRQ
y
3LORWUXQDQG
EHQHILWWUDFNLQJ
y
L9$/8(/HDG
6FRULQJ
y
5ROORXWSODQQLQJ

y
%DVHOLQHH[LW
FULWHULDIRU
L9$/8(PRGHOV

y
'HVLJQIRU:HE
6HUYLFHV
y
L9$/8($WWULWLRQ
6XUYLYDOPRGHO
WUDLQLQJ

y
L9$/8(PRGHO
WHVWLQJDJDLQVW
H[LWFULWHULD

y
'HVLJQIRU
FKDQJH
PDQDJHPHQW
y
7UDLQLQJ

,QVLJKWV
RQD3RUWDO

%XVLQHVV
,QWHJUDWLRQ
$QDO\WLFV
6HUYLFH
,QWHJUDWLRQ

y
&XVWRPL]DWLRQRI y
&XVWRPL]DWLRQRI
L9$/8(SRUWDO
L9$/8(:HE
6HUYLFHVIRU
y
&XVWRPL]DWLRQRI
VFRULQJ
EXVLQHVVXWLOLWLHV
IRUDJHQWV1%
y
,QWHJUDWLRQRI
8:DQGSROLF\
L9$/8(:HE
VHUYLFLQJDQG
6HUYLFHVWR
FDOOFHQWHU
DJHQF\1%8:
RSHUDWLRQV
SROLF\VHUYLFLQJ
DQGFDOOFHQWHU
y
L9$/8(ODXQFK
V\VWHPV
y
,QWHJUDWLRQZLWK
VRXUFHV\VWHPV
IRUGDWDLQWDNH

L9$/8(,PSOHPHQWDWLRQ&DVHVWXG\
6ROXWLRQ

6LWXDWLRQ

&OLHQWH[SHULHQFHGKLJKDWWULWLRQLQ
LWVFXVWRPHUEDVHGXHWRFRPSHWL
WLRQDQGPLVVHOOLQJLQRYHUDOO
ODSVHUDWH
 %XVLQHVVZDVKHDYLO\GHSHQGHQWRQ
RQH/2%LV0HGLFDOFRPSDUHG
WRWKHLQGXVWU\DYHUDJH
 &RPSDQ\ZDVKLJKO\UHJLRQIRFXVHG
DQGQRWDQDWLRQDOLQVXUHU6ROGRQO\
LQRIWKHUHJLRQDOPDUNHWV
 %XVLQHVVQHYHUH[DPLQHGWKH
PRWLYDWLRQIRUFXVWRPHUVWREX\DQG
KRZWKH\FRXOGEHSURDFWLYHO\
UHWDLQHG

 3HUIRUPHGFXVWRPHUVHJPHQWDWLRQ
WRXQGHUVWDQGFXVWRPHU
FKDUDFWHULVWLFV
 &UHDWHGWKHDWWULWLRQPRGHOWR
SUHGLFWSROLF\FKXUQXVLQJ/RJLVWLF
UHJUHVVLRQDQGVXUYLYDOPRGHODQG
FDOFXODWHWKHGXUDWLRQRIDSROLF\
EHIRUHDWWULWLRQ
 'HYHORSHGXWLOLWLHVIRUXQGHUZULWHUV
WRLGHQWLI\SRWHQWLDODWWULWLRQ
SUREDEOHVEHIRUHXQGHUZULWLQJQHZ
EXVLQHVVSURSRVDOV
 'HYHORSHGSURGXFWLYLW\WRROVIRU
DJHQWVWRUHFRPPHQGWKHULJKW
SURGXFWVEDVHGRQSURVSHFWSURILOHV

%HQHILWV
 7ZHQW\ILYHSHUFHQWRIWKHLQIRUFH
SRUWIROLRKDVEHHQLGHQWLILHGZLWK
RIWKHSUHPLXPDWULVNRI
DWWULWLRQ$SSUR[86%1
SUHPLXPDWULVN
 'HYHORSHGWDUJHWOLVWRIFXVWRPHUV
ZKRFDQEX\VSHFLILFSURGXFWV
 *HQHUDWHGOHDGVIURPH[FXVWRPHUV
ODSVHGFXVWRPHUVZRUWK
860LOOLRQZLWKHVWLPDWHG
SUREDELOLW\RIFRQYHUVLRQ
 'HYHORSHGUHFRPPHQGDWLRQVIRU
SUHPLXPUHWHQWLRQIRUWKH
FXVWRPHUUHWHQWLRQDQG
SHUVLVWHQF\PDQDJHPHQWWHDPV
 )RUPXODWHGDJURZWKVWUDWHJ\WR
UHWDLQH[LVWLQJFXVWRPHUVDQGJURZ
WKHEXVLQHVV

)LJXUH

cognizant 20-20 insights

Sales Productivity Analysis

Sales productivity can be measured by tracking
several metrics, each providing a slightly different
perspective on the concept of productivity.
Some commonly used metrics employed across
distribution channels are quote-to-close ratio,
gross written premium (GWP) and per-agent
ratio, which calculates sales and marketing costs
per US$1,000 of placed first-year annualized
premiums.
Beyond the operational-type metrics listed above,
it is also possible to use data-mining techniques
such as profiling to derive insights from those
customer segments that were most productive.
For example, to achieve a minimum threshold
of revenue generation, advanced analytics can
identify the lowest cost associated with taking
a specific customer segment from quote to
close. That is, the cost to usher these customers
through the sales cycle from presentation
of the illustration to completion of the sale.
Similar profiling analyses can also be conducted
with distributors to identify the demographic,
geographical and other factors that drive
successful and efficient business practices.
Profiling can also help make comparisons
between sales territories, and lead to a more
level playing field. Questions such as "If New York
generates 2x the revenue as Michigan, does that
make it twice as good? twice as profitable?
twice as efficient?" can best be addressed
through geographic profiling that enables an

understanding of discrete markets. Again, sales

productivity can be measured in several ways,
and suitable analytical approaches can involve a
blend of operational and predictive techniques.
Agent and Salesforce Deployment Analysis
Agent and salesforce deployment analysis
addresses critical resource-allocation tasks, such
as territory alignment, workload allocation and
salesforce sizing. Historically, these activities
were conducted manually generally resulting in
unbalanced workloads, impractical quotas and a
lack of focus on key accounts. Analytics can help
establish a benchmark to build territories that are
distinctly different and independently managed.
This can enable leadership and individual distribution partners to more precisely measure their
comparative performance. The same benchmarking tools and techniques used at the carrier level
can make it easier for independent distributors
and alternate distribution channels to conduct
fair performance appraisals. Moreover, the use
of this type of platform allows greater transparency throughout the organization, and a level
of insight that enables smaller geographies and
territories to be equitably compared and measured against larger players.
It is possible to mathematically model goals,
quotas, activities and staffing levels to get a
near correct answer to complex questions
surrounding resource deployment. The output
from mathematical models can then be manually
tweaked to work on the ground, given the
marketplace realities that cannot be factored into

Quick Take
How Analytics Helped an Insurer Create a 360-Degree Agent View
A top-five P&C insurance carrier based in North America had an agent attrition problem that was in the
top decile* of the industry. Working with us, the carrier designed machine learning algorithms to study
patterns that existed across internal and external data. The models were used to identify the likelihood
of attrition of the carriers existing agents; develop a real-time profile of them, including information on
agents lifetime value; the time to reach their full potential; likely and pessimistic production forecasts;
plus the reason for and expected time of departure.
Subsequently, these models were used to screen potential recruits, generate a profile and accelerate the
recruitment of the right set of agents.
The insurer was able to increase its sales conversion rates 20% by applying quote conversion analytics
and revisions to its quote follow-up processes, which can be redesigned to run in campaign mode using
insights generated from analytics.
* Decile refers to any of the groups that result when a frequency distribution is divided into ten groups of equal size.

cognizant 20-20 insights

$%OXHSULQWIRU6XFFHVV
%XVLQHVV6LWXDWLRQ

2XU$SSURDFK

7KHFOLHQW
VUHFUXLWPHQWWHDP
ZDQWHGWR
v8QGHUVWDQGWKHVXFFHVVRI
HPSOR\HHVLQWKHPHUFKDQW
OHDGHUVKLSSURJUDP
v([SORUHWKHIDFWRUVWKDWDIIHFW
VXFFHVVJURZWKDQGWHQXUHRI
FDQGLGDWHVDIWHUWUDLQLQJ
v3URDFWLYHO\LGHQWLI\DSHUVRQDO
L]HGWUDLQLQJVFKHGXOHWRSUHSDUH
UHFUXLWVIRUVXFFHVV

v'HILQH.3,VWRGHILQH
VXFFHVVJURZWKDQG
WHQXUHWKHPHV
v$QDO\]HKLVWRULFDOGDWD
DQGYLVXDOL]HGHVFULSWLYH
LQVLJKWV
v(VWDEOLVKVXFFHVVIDFWRUV
OLNHHGXFDWLRQXQLYHUVLW\
HWFXVLQJORJLVWLF
UHJUHVVLRQWRREWDLQ
SUHGLFWHGWUHQG

5HVXOWV
v3UHGLFWHGHPSOR\HHVXFFHVV
JURZWKDQGWHQXUHLQWKH
RUJDQL]DWLRQ
v'HYHORSHGDGDVKERDUGZLWK
RSHUDWLRQDOPHWULFVWLPH
WHQXUHHWFRIWKHQHZ
UHFUXLWVDQGLQWHUQV
v%DVHGRQWKHUHFRPPHQGDWLRQV
WKHUHFUXLWLQJSURFHVVKDVEHHQ
VWUHDPOLQHGE\WDNLQJVXFFHVV
IDFWRUVLQWRDFFRXQW

.H\)HDWXUHV
3UHGLFWLRQRIHPSOR\HHWHQXUH
%DVHGRQHPSOR\HHDWWULEXWHVSUHGLFW
HPSOR\HH
VWHQXUHZLWKWKHRUJDQL]DWLRQ
3DUDPHWHUVDIIHFWLQJVXFFHVV
,GHQWLI\NH\DWWULEXWHVLPSDFWLQJWKH
VXFFHVVRIDQHPSOR\HHVXFKDVHGXFD
WLRQDOEDFNJURXQGDQGGHPRJUDSKLFV
5HFRPPHQGDWLRQRQKRZWRUHGXFH
UHFUXLWPHQWWLPH
(PSOR\DFWLRQDEOHLQVLJKWVWRUHGXFHWKH
UHFUXLWPHQWWLPHRIHPSOR\HHV

)LJXUH

a mathematical model. For example, in salesforce

sizing and allocation situations, nonlinear
programing models that maximize three-to-five
years of profitability for alternative salesforce
sizes (and market allocations) can be used as a
technique for developing first-cut deployment
models. Output from this analysis can then be
fine-tuned, based on personnel strengths and
relationships in certain markets.
For sales-territory designs where the distribution of
territories and accounts is in question, integrated
programming models that maximize coverage and
minimize profit disruptions can typically be used.
Furthermore, as shown in Figure 2 above, analytical
models can be employed to identify factors that
determine the success of newly inducted agents/
brokers.
Incentives and Commissions Analysis
Optimizing payouts on incentives and commissions is critical if a carrier is to maximize its overall profitability and ensure the retention of quality distribution staff. By using the benchmarking
referenced earlier, carriers can better understand

cognizant 20-20 insights

profitability per distributor and therefore can

control it better. One element of this approach is
that distributors are not paid exclusively based
on premium just one factor in a mix that collectively rewards efficient and profitable businesses.
To gain a meaningful perspective on the interplay among these variables, it is also possible to
calculate average payouts per distribution agent
and compare those with industry benchmarks (or
historic values) to see how the current level of
incentive payments measures up when compared
to where it needs to be.
Advanced computational models can be used to
align compensation with the level of sales efforts.
These kind of non-standard models can help
build incentive compensation plans that extend
beyond the considerations of company sales and
profits. For example, disaggregate models link
each salespersons utility for time; money and
sales response models link each salespersons
call efforts with territory sales to develop the
aggregate relationship. While such options are
rarely used, they can help carriers gain a unique,
computationally-intensive perspective on the
efficiency of their sales operation.

Channel Preference Model

As discussed earlier, the modern digital customer
expects a seamless, multi-channel experience
once he or she starts interacting with an insurance
carrier and its distribution network. While
these customers may interact through various
channels, they generally have a preference. It is
possible to collect data and discern patterns on
preferred channels and their contribution to the
carriers top line. Operational intelligence metrics
can gather the data on various touchpoints and
sequence those channels based on customer
interactions.
In addition to operational data, the analytics
team can conduct data mining exercises such as
classification studies to predict which channel
may be preferred by an incoming new lead based
on demographic or behavioral attributes. These
studies can also be used to predict the probability
of conversion or cross-sale, or chances of
responding to certain offers through the available
distribution channels.
Churn Prediction Model
Churn rate is a measure of customer or employee
attrition. Strictly defined, it is the number
of customers that discontinue a service, or
employees who leave a company during a specified
time period, divided by the average total number
of customers or employees over that same time
frame. Customer retention is a critically important
business goal in todays competitive arena. An
in-depth analysis of factors affecting customer
attrition helps in understanding the reasons
behind why loyal customers drift toward other
carriers. These insights can help in formulating
effective strategies that can improve customers
engagement and keep them from turning to
competitors.
Often, carriers are interested in knowing if a
customer will leave when their current policy
lapses. In these situations, predictive churn
analysis can help estimate certain target-level
outputs, such as the probability that a particular
customer segment or archetype will leave, given
a proposed premium rate offer or other changes
in the product offering or price. This analytical
approach involves a supervised classification, or
class probability, estimation. It produces a model
that categorizes all potential customers into
churn and no churn groups if current offer
parameters are retained in the renewal offer.

cognizant 20-20 insights

This can make it very easy for renewal analysts

to target and come back to the high chance of
churn customers with a more competitive offer
addressing attrition before it happens. Making
a revised offer after the consumer has switched
carriers is often too little too late.
Sales Campaign (Including Cross-Sell/
Up-Upsell) Analysis
The ability to cross-sell a product throughout the
business or up-sell another, higher-value product from the same organization is key to driving up net revenue. The
rationale? Selling to cur- For sales-territory
rent customers involves designs where
no additional costs, unlike
selling to new customers. the distribution
Traditionally, the carriers/ of territories and
distributors sales efforts
accounts is in
are aligned by product
type or solution type. question, integrated
While solution-type align- programming
ment can support crossmodels that
sell opportunities, it can
lead to short-term resis- maximize coverage
tance, since some view and minimize profit
solution similarities as a
disruptions can
threat that could cannibaltypically be used.
ize other product sales.
The most effective analytical methods for
overcoming these issues and driving up-sell
and cross-sell revenues are generally those
associated with similarity and clustering
categories. If two attributes (among people,
companies, products) are similar in some ways,
there is a good chance that other characteristics
will be shared as well. Data-mining techniques
that focus on searching for the right sort
of similarity (or degree of association) help
identify the next product or service that a
current customer may be interested in because
a similar customer has already bought it and
likes it. Modern retailers such as Amazon and
Netflix use similarity or degree of association
(i.e., collaborative filtering) to provide product
recommendations a core tenet of Code Halo
thinking (people who liked X also liked Y.).
Such recommendation engines could be used
more pervasively in insurance to add relevancy
a valued attribute for customers to the purchase
and decision-making process for prospects at
the time of the first purchase, as well as for
existing customers whose requirements change
over their lifetime.

Customer Service Analysis

Customer service analysis tracks and investigates
customer interactions with an organization
at point-of-sale, service delivery and during
follow-on support to ensure that customers are
well served. Effective customer service is an
integral part of a successful insurance brand
experience, and therefore closely monitored
by the most senior management. Operational
reports track service levels at customer touch
points, and issue resolution rates and changes in
the number of customer requests and complaints.
Several metrics commonly used for this purpose
include dollars to resolve issue, number of issues
outstanding and CSAT (customer satisfaction)
ratings.
While using analytics to track the behavior
pattern of a dissatisfied customer is a relatively
new approach, it is proving to be highly valuable.
The customer experience is not a binary event. A
not bad customer experience does not mean it
is a good one a fact often
A not bad overlooked by carriers and
customer distributors. Insurers must
understand how analytics
experience does can be used to help design
not mean it is a a service that clients will
good one. clamor for even if they have
minimal knowledge of the
offering. Clients desire a level of comfort, since
purchasing many insurance products involves a
lifetime financial commitment. The goal should be
to move the customer to self-service mode and
reduce the chance of displeasing them.
Using an analytical approach known as profiling,
insurers can examine customer service data
to address questions concerning the typical
behavior patterns of a dissatisfied customer, or
predict whether a particular customer complaint
or claim is just an anomaly or points to a larger
yet hidden trend. Profiling can also be used as
input for data-reduction analytical techniques
(transforming digital information into an orderly,
simplified form) that focus on understanding
events or actions that might lead to more
customer issues.

financial status stay or leave after their policy

lapses? There are many techniques for generating
supervised segmentation from a dataset. One of
the most popular approaches is to create a treestructured model, where different branches of
the tree represent the segments into which the
data can be distributed. Sometimes, the analyst
may be interested in more information than
merely classifying customers into segments. He
or she may want to know the probability of each
event within a class (predicting the probability of
someone leaving within three months compared
with just predicting if he/she will leave within
three months of joining).
Overall, customer segmentation analysis can
give carriers a very useful grasp of the aspects
of consumer behavior, which can be a precursor
to actions such as churn, complaints and fraud.
Timely prevention can translate into huge savings
in what are sometimes considered unavoidable
distribution losses.
Customer Profitability Analysis
For the purposes of this paper, we can assume that
the average customer of a distribution company
also happens to be a profitable one. That alone
does not make them an ideal customer. The ideal
customer clears all or nearly all of the internal
benchmarks related to profitability and retention.
Many businesses have defined this quite well
often attracting a lot of ideal customers by
analytically aligning the organization to find them.
Therefore, to the extent that there is an unending
supply of these ideal customers, a business
would want to work exclusively with these clients.
The reality is that there is not an unending supply
of ideal candidates at least not all the time.
Businesses would be well advised to consider the
channelization of prospective customers. In this
approach, it is understood that a blended average
across the migration continuum is a success
if it meets some internal standards.

6WURQJ
$YHUDJH

Customer Segmentation
This form of analysis addresses fundamental
questions such as How can we select one or
more attributes/features/variables (like churn
rate) that will best divide the sample with respect
to the target variable of interest? For example,
will the customer of a certain demographic and
cognizant 20-20 insights

,GHDO
)LJXUH

10

$FFHSWDEOH
1RW$FFHSWDEOH

Clearly, the goal would be to obtain as many

profitable customers at the top end of the continuum
while achieving minimum thresholds of client
profitability. Profitability analysis would then mean
calculating differences between highly profitable
and less profitable client segments along this
continuum. Questions of this kind can be answered
through the use of statistical techniques, such as
hypothesis testing.
At the same time, insurance distribution managers
are often interested in answering more intriguing
questions, such as Who are really our most profitable
customers? Is there a way to characterize them? Are
there any character attributes that lend themselves
well to lowest-cost distribution channels? They also
want to know early on if a particular new customer
will be a profitable one and how much revenue
they should expect that customer to generate.
Understanding true customer profitability can
change, based on the perspective used to examine
the data. Various analytical methods that fall into
classification and regression analysis categories
can be used in conjunction with creative business
analysis to develop a well-rounded and balanced
understanding of customer profitability.

Looking Forward
Insurance and the entire financial services
industry are at a crossroads as established
players wrestle with a host of challenges and
opportunities. Advanced analytic capabilities hold
enormous potential for leaders to evaluate their
business and allow it to thrive. Understanding the
evolving roles of the carrier, producing agents,
distribution structures and digitally connected
consumers will distinguish the progress carriers
make moving forward.
Winners will view these factors through the lens
of opportunity, and harness the power of smart
analytics to benefit their clients, strengthen their
profit margins and establish a durable competitive
position.
We believe that an integrated analytics strategy
will provide carriers and distribution channels
with an added advantage by enabling them to:

Better serve the client (i.e., the digital

customer) by gaining a clearer understanding of what they want and how they want to
be engaged. Today, carriers can create tighter

Predicting Outperformance
Solution Approach

Targeted
Training
Recommendations

tage of
Percen
ees
lo
p
em y
nt time
ployme
per de

Deployment Time

Parameters
Affecting
Deployment Time

Educational
Background
Recommendation

80%

4,946

60%

40%
1,744

2
547
0

20%
112

Performance

Figure 4

1.5 2
Years

1 1.5
Years

6 months
1 Year

36
Months

Less than
3 Days

50%
40%
30%
20%
10%
0%

0%

Percentage of
employees per
performance

Key
Features

10
8
6
4
2
0

Count of Employees
(000s)

Evaluate the factors

impacting the
performance of an
individual after
induction training.

% of employees (000s)

Historical data review

Logistic models to
compute the probability of becoming fully
productive at various
points in time
Survival models to
predict the ability of
the employee to
sustain his or her
productivity levels

Estimate the time

new employees take
to become fully
productive after they
join the organization.

Deployment
Time
Prediction

Results

Less then
0 Days

Business Problem

continued on next page...

cognizant 20-20 insights

11

77%

High

City#2

76%

High

Target

Discipline#2 Training#1

City#2

73%

High

Target

Discipline#1

Training#2

City#1

57%

Medium

NonTarget

Discipline#1

Training#1

City#1

56%

Medium

Target

Discipline#1

Training#2

City#2

55%

Medium

Target

Discipline#2 Training#2

City#2

51%

Medium

NonTarget

Discipline#2 Training#1

City#2

50%

Low

NonTarget

Discipline#1

Training#2

City#1

33%

Low

NonTarget

Discipline#2 Training#2

City#1

30%

Low

NonTarget

Discipline#2 Training#2

City#2

28%

Low

Probability
Status

Probability
Status

City#1

Training#1

Probability
of Rtings
A&M

Deployment
Probability
(within three
months)

Training#1

Discipline#1

Training
Group

Training
Location
Group

Discipline#1

Target

Dovetail
Score

Training
Group

Target

College
Group

Discipline
Group

Performance Matrix: Probability of

Employee Performing Well (Above Average)

College
Group

Deployment Matrix: Probability of getting deployed

in less than three months

Target

>76.663

Training
#1

87%

High

Target

>76.663

Training
#2

81%

High

Target <=76.663

Training
#1

77%

High

NonTarget

<=72.5

Training
#1

72%

Medium

NonTarget

>=60

Training
#2

69%

Low

Stakeholders received a priority matrix table of colleges for recruitment and

specialized training, thereby knowing what combinations work best.

Figure 4

connections and a stronger brand proposition

with customers. For example, they can leverage the snippets from customer Code Halos
to create a Personal Digital Portal where a
client can manage his application, policy and
future preferences. This portal could also
allow policyholders to receive tailored company news feeds and provide customer-survey
feedback to enhance product innovations.

Provide systematic guidance and a roadmap for sales and distribution by facilitating
well informed planning and sales activities
throughout the organizational hierarchy
from senior leadership to the producing agent.

Create well planned, informative sales and

marketing campaigns that support superior
results for agents. Campaigns can be more

cognizant 20-20 insights

customized as carriers distill insights from

customers Code Halos.

Tackle the challenges of field management

and producing agents related to territory
development, lead creation/management and
performance tracking across the organization.

Just as other industries have transformed their

operations through the smart use of digital
technologies, so must the insurance sector. In
the end, the integration of analytics will drive
more simplicity and greater efficiencies across
the industry and within individual businesses.
This transformation will not only deliver more
value to the business and internal stakeholders,
but also, most critically, create a compelling,
satisfying and lasting experience for todays
digitally-minded consumers.

12

Footnotes
Guruveer Vasir. Will Bancassurance Become the Major Distribution Channel for Insurance? Industry
ReportStore.Com. http://industryreportstore.blogspot.com/2014/06/will-bancassurance-becomemajor.html.

For more on Code Halos and innovation, read Code Rules: A Playbook for Managing at the
Crossroads. Cognizant Technology Solutions, June 2013, http://www.cognizant.com/Futureofwork/
Documents/code-rules.pdf, and the book, Code Halos: How the Digital Lives of People, Things,
and Organizations are Changing the Rules of Business, by Malcolm Frank, Paul Roehrig and Ben
Pring. Published by John Wiley & Sons, April 2014. http://www.wiley.com/WileyCDA/WileyTitle/
productCd-1118862074.html.

Sathyanarayanan Sethuraman. Can Amazon Dominate in Insurance, Too?

InsuranceThoughtleadership.com. http://insurancethoughtleadership.com/can-amazon-dominate-ininsurance-too/.

Analytics and Mobile Top Priority Technologies for Insurers This Year. Strategy Meets Action
(SMA) Report. https://strategymeetsaction.com/our-research/2014-insurance-ecosystem-insurertechnology-spending-drivers-and-projects.

References

Analytics and Mobile Top Priority Technologies for Insurers This Year: Strategy Meets Action (SMA)
Report. http://www.propertycasualty360.com/2014/01/28/analytics-mobile-are-top-tech-prioritiesfor-insur.

Three Quarters of Insurance Companies Use Predictive Analytics: ISO Survey.

http://www.insurancetech.com/three-quarters-of-insurance-companies-use-predictive-analytics-inpricing-iso-survey/d/d-id/1314850.

Data Science for Business: What You Need to Know About Data Mining, Second Edition, 2013.
Provost, Foster, and Tom Fawcett. OReilly Media.

James McQuivey. Digital Disruption: Unleashing the Next Wave of Innovation, First Edition.
Copyright Forrester Research, Inc. 2013. Amazon Publishing.

Acknowledgments
The authors would like to thank Matt Hamann, an industry expert on the digital customer experience,
for his contributions to this white paper.

About the Authors

Chandan Gokhale is a Senior Director with Cognizants Insurance Business Unit. In this role,
he is responsible for growing the units digital transformation sub-practice and lending support to
Cognizants R&D arm, InsuranceNext Labs. He is a qualified design strategist and TOGAF-certified
systems architect, with more than fifteen years of consulting experience with some of the top technology
and design firms in the industry. He has strong experience in leading design-strategy and technologyinnovation co-creation engagements. He can be reached at Chandan.Gokhale@cognizant.com.
Ajish Gopan is the Head of Cognizants North American Insurance Analytics Practice, which serves a
range of insurance carriers across the Life and P&C spectrum. Ajish is an alumnus of IIT Madras and IIM
Bangalore. He is also a LOMA-certified insurance professional, and has completed an executive program
from Harvard University. He can be reached at Ajish.Gopan@cognizant.com.

cognizant 20-20 insights

13

About Cognizant
Cognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process outsourcing services, dedicated to helping the world's leading companies build stronger businesses. Headquartered
in Teaneck, New Jersey (U.S.), Cognizant combines a passion for client satisfaction, technology innovation, deep
industry and business process expertise, and a global, collaborative workforce that embodies the future of work. With
over 75 development and delivery centers worldwide and approximately 211,500 employees as of December 31, 2014,
Cognizant is a member of the NASDAQ-100, the S&P 500, the Forbes Global 2000, and the Fortune 500 and is ranked
among the top performing and fastest growing companies in the world.
Visit us online at www.cognizant.com or follow us on Twitter: Cognizant.

World Headquarters

European Headquarters

India Operations Headquarters

500 Frank W. Burr Blvd.

Teaneck, NJ 07666 USA
Phone: +1 201 801 0233
Fax: +1 201 801 0243
Toll Free: +1 888 937 3277
Email: inquiry@cognizant.com

1 Kingdom Street
Paddington Central
London W2 6BD
Phone: +44 (0) 207 297 7600
Fax: +44 (0) 207 121 0102
Email: infouk@cognizant.com

#5/535, Old Mahabalipuram Road

Okkiyam Pettai, Thoraipakkam
Chennai, 600 096 India
Phone: +91 (0) 44 4209 6000
Fax: +91 (0) 44 4209 6060
Email: inquiryindia@cognizant.com

Copyright 2015, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is
subject to change without notice. All other trademarks mentioned herein are the property of their respective owners.